man7/cgroup_namespaces.7

   1 .\" Copyright (c) 2016 by Michael Kerrisk <mtk.manpages@gmail.com>
   2 .\"
   3 .\" %%%LICENSE_START(VERBATIM)
   4 .\" Permission is granted to make and distribute verbatim copies of this
   5 .\" manual provided the copyright notice and this permission notice are
   6 .\" preserved on all copies.
   7 .\"
   8 .\" Permission is granted to copy and distribute modified versions of this
   9 .\" manual under the conditions for verbatim copying, provided that the
  10 .\" entire resulting derived work is distributed under the terms of a
  11 .\" permission notice identical to this one.
  12 .\"
  13 .\" Since the Linux kernel and libraries are constantly changing, this
  14 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  15 .\" responsibility for errors or omissions, or for damages resulting from
  16 .\" the use of the information contained herein.  The author(s) may not
  17 .\" have taken the same level of care in the production of this manual,
  18 .\" which is licensed free of charge, as they might when working
  19 .\" professionally.
  20 .\"
  21 .\" Formatted or processed versions of this manual, if unaccompanied by
  22 .\" the source, must acknowledge the copyright and authors of this work.
  23 .\" %%%LICENSE_END
  24 .\"
  25 .\"
  26 .TH CGROUP_NAMESPACES 7 2016-07-17 "Linux" "Linux Programmer's Manual"
  27 .SH NAME
  28 cgroup_namespaces \- overview of Linux cgroup namespaces
  29 .SH DESCRIPTION
  30 For an overview of namespaces, see
  31 .BR namespaces (7).
  32 .PP
  33 Cgroup namespaces virtualize the view of a process's cgroups (see
  34 .BR cgroups (7))
  35 as seen via
  36 .IR /proc/[pid]/cgroup
  37 and
  38 .IR /proc/[pid]/mountinfo .
  39 .PP
  40 Each cgroup namespace has its own set of cgroup root directories.
  41 These root directories are the base points for the relative
  42 locations displayed in the corresponding records in the
  43 .IR /proc/[pid]/cgroup
  44 file.
  45 When a process creates a new cgroup namespace using
  46 .BR clone (2)
  47 or
  48 .BR unshare (2)
  49 with the
  50 .BR CLONE_NEWCGROUP
  51 flag, it enters a new cgroup namespace in which its current
  52 cgroups directories become the cgroup root directories
  53 of the new namespace.
  54 (This applies both for the cgroups version 1 hierarchies
  55 and the cgroups version 2 unified hierarchy.)
  56 .PP
  57 When viewing
  58 .IR /proc/[pid]/cgroup ,
  59 the pathname shown in the third field of each record will be
  60 relative to the reading process's root directory
  61 for the corresponding cgroup hierarchy.
  62 If the cgroup directory of the target process lies outside
  63 the root directory of the reading process's cgroup namespace,
  64 then the pathname will show
  65 .I ../
  66 entries for each ancestor level in the cgroup hierarchy.
  67 .PP
  68 The following shell session demonstrates the effect of creating
  69 a new cgroup namespace.
  70 First, (as superuser) we create a child cgroup in the
  71 .I freezer
  72 hierarchy, and put the shell into that cgroup:
  73 .PP
  74 .EX
  75 .nf
  76 .in +4n
  77 # \fBmkdir \-p /sys/fs/cgroup/freezer/sub\fP
  78 # \fBecho $$\fP                      # Show PID of this shell
  79 30655
  80 # \fBsh \-c \(aqecho 30655 > /sys/fs/cgroup/freezer/sub/cgroup.procs\(aq\fP
  81 # \fBcat /proc/self/cgroup | grep freezer\fP
  82 7:freezer:/sub
  83 .in
  84 .fi
  85 .EE
  86 .PP
  87 Next, we use
  88 .BR unshare (1)
  89 to create a process running a new shell in new cgroup and mount namespaces:
  90 .PP
  91 .nf
  92 .EX
  93 .in +4n
  94 # \fBunshare \-Cm bash\fP
  95 .in
  96 .EE
  97 .fi
  98 .PP
  99 We then inspect the
 100 .IR /proc/[pid]/cgroup
 101 files of, respectively, the new shell process started by the
 102 .BR unshare (1)
 103 command, a process that is in the original cgroup namespace
 104 .RI ( init ,
 105 with PID 1), and a process in a sibling cgroup
 106 .RI ( sub2 ):
 107 .PP
 108 .nf
 109 .EX
 110 .in +4n
 111 $ \fBcat /proc/self/cgroup | grep freezer\fP
 112 7:freezer:/
 113 $ \fBcat /proc/1/cgroup | grep freezer\fP
 114 7:freezer:/..
 115 $ \fBcat /proc/20124/cgroup | grep freezer\fP
 116 7:freezer:/../sub2
 117 .in
 118 .EE
 119 .fi
 120 .PP
 121 From the output of the first command,
 122 we see that the freezer cgroup membership of the new shell
 123 (which is in the same cgroup as the initial shell)
 124 is shown defined relative to the freezer cgroup root directory
 125 that was established when the new cgroup namespace was created.
 126 (In absolute terms,
 127 the new shell is in the
 128 .I /sub
 129 freezer cgroup,
 130 and the root directory of the freezer cgroup hierarchy
 131 in the new cgroup namespace is also
 132 .IR /sub .
 133 Thus, the new shell's cgroup membership is displayed as \(aq/\(aq.)
 134 .PP
 135 However, when we look in
 136 .IR /proc/self/mountinfo
 137 we see the following anomaly:
 138 .PP
 139 .nf
 140 .EX
 141 .in +4n
 142 # \fBcat /proc/self/mountinfo | grep freezer\fP
 143 155 145 0:32 /.. /sys/fs/cgroup/freezer ...
 144 .in
 145 .EE
 146 .fi
 147 .PP
 148 The fourth field of this line
 149 .RI ( /.. )
 150 should show the
 151 directory in the cgroup filesystem which forms the root of this mount.
 152 Since by the definition of cgroup namespaces, the process's current
 153 freezer cgroup directory became its root freezer cgroup directory,
 154 we should see \(aq/\(aq in this field.
 155 The problem here is that we are seeing a mount entry for the cgroup
 156 filesystem corresponding to our initial shell process's cgroup namespace
 157 (whose cgroup filesystem is indeed rooted in the parent directory of
 158 .IR sub ).
 159 We need to remount the freezer cgroup filesystem
 160 inside this cgroup namespace, after which we see the expected results:
 161 .PP
 162 .nf
 163 .EX
 164 .in +4n
 165 # \fBmount \-\-make\-rslave /\fP     # Don't propagate mount events
 166                             # to other namespaces
 167 # \fBumount /sys/fs/cgroup/freezer\fP
 168 # \fBmount \-t cgroup \-o freezer freezer /sys/fs/cgroup/freezer\fP
 169 # \fBcat /proc/self/mountinfo | grep freezer\fP
 170 155 145 0:32 / /sys/fs/cgroup/freezer rw,relatime ...
 171 .in
 172 .EE
 173 .fi
 174 .PP
 175 Use of cgroup namespaces requires a kernel that is configured with the
 176 .B CONFIG_CGROUPS
 177 option.
 178 .\"
 179 .SH CONFORMING TO
 180 Namespaces are a Linux-specific feature.
 181 .SH NOTES
 182 Among the purposes served by the
 183 virtualization provided by cgroup namespaces are the following:
 184 .IP * 2
 185 It prevents information leaks whereby cgroup directory paths outside of
 186 a container would otherwise be visible to processes in the container.
 187 Such leakages could, for example,
 188 reveal information about the container framework
 189 to containerized applications.
 190 .IP *
 191 It eases tasks such as container migration.
 192 The virtualization provided by cgroup namespaces
 193 allows containers to be isolated from knowledge of
 194 the pathnames of ancestor cgroups.
 195 Without such isolation, the full cgroup pathnames (displayed in
 196 .IR /proc/self/cgroups )
 197 would need to be replicated on the target system when migrating a container;
 198 those pathnames would also need to be unique,
 199 so that they don't conflict with other pathnames on the target system.
 200 .IP *
 201 It allows better confinement of containerized processes,
 202 because it is possible to mount the container's cgroup filesystems such that
 203 the container processes can't gain access to ancestor cgroup directories.
 204 Consider, for example, the following scenario:
 205 .RS 4
 206 .IP \(bu 2
 207 We have a cgroup directory,
 208 .IR /cg/1 ,
 209 that is owned by user ID 9000.
 210 .IP \(bu
 211 We have a process,
 212 .IR X ,
 213 also owned by user ID 9000,
 214 that is namespaced under the cgroup
 215 .IR /cg/1/2
 216 (i.e.,
 217 .I X
 218 was placed in a new cgroup namespace via
 219 .BR clone (2)
 220 or
 221 .BR unshare (2)
 222 with the
 223 .BR CLONE_NEWCGROUP
 224 flag).
 225 .RE
 226 .IP
 227 In the absence of cgroup namespacing, because the cgroup directory
 228 .IR /cg/1
 229 is owned (and writable) by UID 9000 and process
 230 .I X
 231 is also owned by user ID 9000, then process
 232 .I X
 233 would be able to modify the contents of cgroups files
 234 (i.e., change cgroup settings) not only in
 235 .IR /cg/1/2
 236 but also in the ancestor cgroup directory
 237 .IR /cg/1 .
 238 Namespacing process
 239 .IR X
 240 under the cgroup directory
 241 .IR /cg/1/2 ,
 242 in combination with suitable mount operations
 243 for the cgroup filesystem (as shown above),
 244 prevents it modifying files in
 245 .IR /cg/1 ,
 246 since it cannot even see the contents of that directory
 247 (or of further removed cgroup ancestor directories).
 248 Combined with correct enforcement of hierarchical limits,
 249 this prevents process
 250 .I X
 251 from escaping the limits imposed by ancestor cgroups.
 252 .SH SEE ALSO
 253 .BR unshare (1),
 254 .BR clone (2),
 255 .BR setns (2),
 256 .BR unshare (2),
 257 .BR proc (5),
 258 .BR cgroups (7),
 259 .BR credentials (7),
 260 .BR namespaces (7),
 261 .BR user_namespaces (7)